Pushkar Jajoria

About Me

I am a PhD at UdS, where I work on applying language models to chemistry. Before this, I worked on computational musicology and generative models for music with latent diffusion models at the University of Galway. Further back, I spent several years in industry as a software engineer at companies like Salesforce and Delhivery, before making a deliberate move toward research.

Outside academia, I am a guitarist and movie enthusiast.

Now a days, I am broadly curious about the cognitive and computational foundations of intelligence, reading whether minds are truly something new that emerges from scale and data, or whether something essential is still missing.

Research Interests

My work sits at the intersection of machine learning and chemistry.

Right now, I’m especially interested in how language models can help us understand molecules. Some of the things I work on:

• Predicting molecular properties using chemical language models
• Testing what large language models really understand about chemistry
• Improving how these models reason about chemical problems
• Exploring in-context learning for chemistry tasks

More broadly, I’m trying to understand the limits of this approach. How far can we push chemistry through language modeling? And where does it start to break completely?

If you’re interested in this space–maybe as a thesis student or collaborator–feel free to reach out.

Publications

JAJORIA, P., GRACIOTTI, A., CASALI, G., ALABI, J. O., DELMONTE, R., POMPILIO, A., TRIPODI, R., MCDERMOTT, J., & KLAKOW, D. (2026). Audio-Lyrics Alignment Dataset for Italian Arias. Proceedings of the LREC 2026.

Shahid, A., Sweeney, R., Jajoria, P., Diamond, D., D’Aquin, M., & McDermott, J. (2025). Ontologies, knowledge graphs, and user interfaces for exploration of Irish traditional music.

JAJORIA, P., KLAKOW, D., & MCDERMOTT, J. (2025). TEXT CONDITIONED SYMBOLIC DRUMBEAT GENERATION USING LATENT DIFFUSION MODELS. Proceedings of the 22nd Sound and Music Computing Conference (SMC2025), Graz, July 2025, 311–318. https://doi.org/10.5281/zenodo.15838008

SWEENEY, R., JAJORIA, P., DIAMOND, D., D’AQUIN, M., & MCDERMOTT, J. (2024). PATTERNS UI, AN INTERACTIVE TOOL FOR MUSIC EXPLORATION. Perspectives on immersion through laser doppler vibrometry, 65.

JAJORIA, P., MADER, A., KRENN, F., & MCDERMOTT, J. (2024). PREVALENCE OF TRESILLO RHYTHM IN CONTEMPORARY POPULAR MUSIC. In Sound and Music Computing.

Jajoria, P. (2021). Sequential prediction of polyphonic music using interpretable features.. (Masters dissertation, Universita della Svizzera Italiana).